Tunnel water inrush disaster frame difference method automatic identification and early warning method and system

Abstract

The invention belongs to the technical field of tunnels and underground engineering. According to the tunnel water inrush disaster frame difference method automatic identification early warning method and system, the number of water inrush points, the water inrush amount and the water inrush quality are monitored in real time, early warning is carried out, and the defect that tunnel water inrush disaster monitoring is carried out manually is overcome; the problems of poor tunnel environment, low monitoring video precision and poor real-time performance in manual monitoring are solved, the labor cost is reduced, and the construction safety is more effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnels and underground engineering, and in particular to a frame difference method and system for automatic identification and early warning of tunnel water inrush disasters. Background Art

[0002] The statements in this section merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] Due to the relatively complex groundwater environment in karst areas, the current construction of tunnels and underground projects is facing construction difficulties such as strong karst, high water pressure, deep burial, high stress, and long tunnel lines. Among them, sudden water disasters have become an extremely severe geological disaster that affects the construction of tunnels and underground projects. Tunnel sudden water disasters are often highly concealed and highly destructive. Once a disaster occurs, it can flood the tunnel and interrupt construction at the least, or cause heavy casualties and economic losses, and the project is forced to stop or reroute, bringing severe challenges to the safe construction of the project. Therefore, monitoring and early warning of sudden water during tunnel excavation has become an indispensable part.

[0004] However, the basic operation of traditional video monitoring of sudden water inrush is still mainly based on manual monitoring, and the environment is very dangerous. In addition, due to many hardware limitations, the monitoring images are sometimes not clear, especially when there are many construction workers and vehicles coming and going during tunnel construction. In some cases, effective information may not be obtained. At this time, relying solely on video monitoring personnel to identify monitoring images or videos and make corresponding judgments will bring huge challenges to the real-time and accuracy of sudden water inrush information. Summary of the invention

[0005] In order to address the deficiencies in the prior art, the present invention provides a method and system for automatic identification and early warning of tunnel water inrush disasters using a frame difference method, which monitors the number of water gushing points, the amount of water inrush and the quality of water inrush in real time, making up for the shortcomings of manual monitoring of tunnel water inrush disasters, and solving the problems of poor tunnel environment, low monitoring video accuracy and poor real-time performance in manual monitoring, which not only reduces labor costs but also more effectively ensures construction safety.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for automatic identification and early warning of tunnel water inrush disasters using a frame difference method.

[0008] A frame difference method for automatic identification and early warning of tunnel water inrush disasters, comprising the following process:

[0009] Acquire images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binary image, perform connectivity processing on the binary image to obtain a target image of a moving target after excluding intrusion by workers or mobile equipment, and thereby obtain a target image sequence in the video;

[0010] The number of water gushing points at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change in the number of water gushing points is predicted using a neural network model based on time series. When the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, an early warning of an increase in the number of water gushing points is issued.

[0011] As a further limitation of the first aspect of the present invention, according to the target image, the contour and area of ​​the sudden water surge are integrally calculated, the amount of the sudden water surge is estimated based on the calculated contour and area of ​​the sudden water surge, and the sudden water surge flow is equivalent to a cylinder to obtain the sudden water surge volume;

[0012] The volume of sudden water surge at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change of the sudden water surge volume is predicted by using a neural network model based on time series. When the sudden water surge volume at a certain moment after the current moment is greater than a second set threshold, an early warning of the increase in the sudden water surge volume is issued.

[0013] As a further limitation of the first aspect of the present invention, based on the target image, an image of a water gushing area is extracted, color information of the water gushing area in each video frame is extracted, the color information extracted in each frame is compared with the color information of a reference frame, and a color difference metric is calculated;

[0014] The color difference measurement of the current moment and multiple moments before the current moment of each video frame is obtained, and the change of the color difference measurement is predicted using a neural network model based on time series. When the color difference measurement at a certain moment after the current moment is greater than a third set threshold, an early warning of water quality change is issued.

[0015] As a further limitation of the first aspect of the present invention, the number of water gushing points, the volume of sudden water gushing, and the change in water quality are used for comprehensive early warning, and the comprehensive early warning score S is calculated, including:

[0016] S=f 1 *ω 1 +f 2 *ω 2 +f 3 *ω 3 , where ω 1 ,ω 2 and ω 3 are weight parameters, f 1 is the first set threshold, f2 is the second set threshold, f 3 Setting a threshold for the third;

[0017] The comprehensive warning index S of the target image sequence at the current moment and multiple moments before the current moment is obtained, and the comprehensive warning score S at a certain moment after the current moment is predicted using a neural network model based on time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing and high risk of water gushing.

[0018] In a second aspect, the present invention provides a frame difference method automatic identification and early warning system for tunnel water inrush disasters.

[0019] A tunnel water inrush disaster automatic identification and early warning system using a frame difference method, comprising:

[0020] The image processing unit is configured to: obtain images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binarized image, perform connectivity processing on the binarized image to obtain a target image of a moving target after excluding intrusion by a worker or a mobile device, and thereby obtain a target image sequence in the video;

[0021] The water gushing point quantity warning unit is configured to: obtain the number of water gushing points at the current moment of the target image sequence and multiple moments before the current moment, use a neural network model based on time series to predict the change in the number of water gushing points, and when the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, issue a warning of an increase in the number of water gushing points.

[0022] As a further limitation of the second aspect of the present invention, the water gushing volume warning unit is configured to: integrally calculate the contour and area of ​​the sudden water gushing according to the target image, estimate the amount of water gushing based on the calculated contour and area of ​​the sudden water gushing, equate the gushing water flow to a cylinder to obtain the sudden water gushing volume, obtain the sudden water gushing volume at the current moment of the target image sequence and multiple moments before the current moment, use a neural network model based on a time series to predict the change of the sudden water gushing volume, and when the sudden water gushing volume at a certain moment after the current moment is greater than a second set threshold, issue a warning of an increase in the sudden water gushing volume;

[0023] The gushing water quality early warning unit is configured to: extract a gushing water area image according to the target image, extract color information of the gushing water area in each video frame, compare the color information extracted in each frame with the color information of the reference frame, calculate the color difference metric, obtain the color difference metric of the current moment and multiple moments before the current moment of each video frame, use a neural network model based on time series to predict the change of the color difference metric, and issue an early warning of water quality change when the color difference metric at a certain moment after the current moment is greater than a third set threshold;

[0024] The comprehensive warning unit is configured to: conduct comprehensive warning based on the number of water inrush points, sudden water volume and water quality changes, and calculate the comprehensive warning score S, including: S = f 1 *ω 1 +f 2 *ω 2 +f 3 *ω 3 , where ω 1 ,ω 2 and ω 3 are weight parameters, f 1 is the first set threshold, f 2 is the second set threshold, f 3 A threshold is set for the third time, and a comprehensive warning index S of the current moment and multiple moments before the current moment of the target image sequence is obtained. The comprehensive warning score S is predicted using a neural network model based on a time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing, and high risk of water gushing.

[0025] In a third aspect, the present invention provides an automatic identification and early warning system for tunnel water inrush disasters using a frame difference method.

[0026] A tunnel water inrush disaster automatic identification and early warning system using a frame difference method, comprising:

[0027] A high-definition camera and a control terminal communicating with the high-definition camera, wherein the control terminal is configured to execute the automatic identification and early warning method for tunnel water inrush disasters using a frame difference method according to the first aspect of the present invention;

[0028] The high-definition camera comprises: a high-definition camera body, a high-definition camera top cover, an inclined lens protection lens, a camera lens, an array-type infrared fill light board, a camera angle adjustment mechanism, a camera mounting bottom plate and a heating wire;

[0029] The top cover of the HD camera is fixed on the top of the HD camera body, the inclined lens protection glass is arranged on the outside of the camera lens, the lens protection glass and the camera lens are arranged at the lower part of the top cover of the HD camera, an array-type infrared fill light board is arranged next to the camera lens, the HD camera body is connected to the camera mounting base plate through a camera angle adjustment mechanism, and heating wires are arranged around the lens protection glass.

[0030] In a fourth aspect, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium;

[0031] a processor adapted to execute a computer program;

[0032] A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by the processor, the automatic identification and early warning method for tunnel water inrush disasters using the frame difference method as described in the first aspect of the present invention is implemented.

[0033] In a fifth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program is suitable for being loaded by a processor and executing the automatic identification and early warning method for tunnel water inrush disasters using the frame difference method as described in the first aspect of the present invention.

[0034] In a sixth aspect, the present invention provides a computer program product, characterized in that the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the automatic identification and early warning method of tunnel water inrush disasters using the frame difference method as described in the first aspect of the present invention.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The corresponding equipment of the present invention collects information such as the water gushing points, water gushing volume, and water quality in the tunnel, analyzes and processes it through machine vision methods, determines whether a sudden water gushing disaster has occurred in the tunnel, as well as information such as the magnitude of the disaster, and comprehensively determines different degrees of early warning, which greatly reduces construction risks, provides a strong basis and guarantee for safe construction, and effectively avoids sudden water gushing disasters during tunnel construction.

[0037] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0039] Figure 1 A flowchart of the automatic identification and early warning method for tunnel water inrush disasters using the frame difference method provided in Example 1 of the present invention;

[0040] Figure 2 A front view of a high-definition camera provided in Embodiment 1 of the present invention;

[0041] Figure 3 A side view of a high-definition camera provided in Embodiment 1 of the present invention;

[0042] Figure 4 A schematic diagram of the installation of a high-definition camera in a tunnel provided in Example 1 of the present invention;

[0043] Figure 5 A schematic diagram of a tunnel water inrush disaster automatic identification and early warning system using a frame difference method provided in Example 2 of the present invention;

[0044] Figure 6 A schematic diagram of a computer device provided in Embodiment 3 of the present invention;

[0045] Among them, 1. HD camera top cover; 2. Lens protection lens; 3. High-performance camera lens; 4. Array infrared fill light board; 5. Camera angle adjustment device; 6. Camera mounting base plate; 7. Heating wire; 8. HD camera body. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0047] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0048] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0049] Embodiment 1:

[0050] As introduced in the background technology, water inrush disasters have become one of the most harmful geological disasters in the construction of tunnels and underground projects due to their strong suddenness, high risk and unpredictability. Once they occur, they often cause significant casualties and property losses. Therefore, during the tunnel construction process, monitoring and early warning of tunnel water inrush disasters has become an effective means and important way to achieve active prevention and control of water inrush disasters and avoid loss of life and property.

[0051] In order to solve the above-mentioned engineering problems, the present invention proposes an automatic identification and early warning method for tunnel water inrush disasters using the frame difference method. The automatic identification and early warning method is based on machine vision and deep learning, and adopts the inter-frame difference method, which is one of the methods of moving target video surveillance detection and tracking. It is implemented through a python algorithm based on the OpenCV library to achieve the monitoring effect of automatic identification and early warning of sudden water inrush disasters in tunnels and underground projects.

[0052] In the automatic recognition and early warning method based on the frame difference method of this implementation, the video file collected by the high-definition camera in the tunnel is converted into a grayscale image in real time, and frame by frame processing is performed to collect the previous frame image and the current frame image. After the grayscale processing, the grayscale value of the corresponding pixel points of two adjacent frames is recorded as f n (x,y) and f n-1 (x, y), subtract the grayscale values ​​of the corresponding pixels of the two frames of images, and take their absolute values ​​to obtain the difference image D n (x, y):

[0053] D n (x, y) = |f n (x, y)-f n-1 (x, y)| (1);

[0054] The static objects are all zero in the difference image, while the moving objects, especially the contours of the moving objects, are non-zero due to the grayscale changes. Preferably, the corresponding threshold T is set, and the pixels in the image are binarized one by one to obtain the binary image R′ n (x, y), where the point with a gray value of 255 is the foreground (moving target) point, that is, the sudden water is detected, and the point with a gray value of 0 is the background point; for image R′ n (x, y) is connected and finally an image R containing a complete moving target is obtained. n .

[0055]

[0056] By judging whether the absolute value of the grayscale difference of each pixel between two frames in the video is greater than the set threshold, the motion characteristics of the video or image sequence are analyzed to determine whether there is any object movement in the image sequence, thereby realizing the monitoring function for moving targets in the surveillance video.

[0057] In this implementation, it is preferred to eliminate the interference caused by construction personnel or equipment during the construction of tunnels or underground projects, and determine whether it is a sudden water intrusion. Specifically, determine whether it is an intrusion by workers or mobile equipment: Since the height and width of a normal person are in a certain ratio, a relatively simple criterion based on this feature is: to judge people by the slope of the motion area, assuming that people in the channel can only walk towards the camera, according to experience: when this slope is less than 0.48, it can be determined as a human intrusion, and greater than 0.48 is non-personnel intrusion. Construction vehicles and other equipment are large in size, and the edge frame size in edge detection is limited to avoid misidentification, which is set to the edge of the moving object being less than a corresponding certain value.

[0058] In this implementation, preferably, the binary image R′ n (x, y) is used for edge detection. Specifically, the image obtained after image binarization is composed of 0 or 1. Therefore, the highest point, the lowest point, the leftmost point, and the rightmost point of the image are found to determine the edge of the moving water burst. The above steps can obtain the outline of the edge of the water burst target, thereby realizing the monitoring function of the water burst.

[0059] After determining that the moving object in the video is a sudden water surge, a judgment standard for sudden water surges was established: the number of water surge points, the amount of water surge, and the identification algorithm of the water quality of sudden water surges, and early warning indicators for determining the level of sudden water surge disasters were set.

[0060] Early warning of the number of water inrush points, including: whether there is water inrush in the video and the change in the number of water inrush points based on the inter-frame difference, on this basis, machine learning processing and early warning release, by judging whether the absolute value of the grayscale difference of each pixel point between two frames in the video is greater than the set threshold to analyze the motion characteristics of the video or image sequence, to determine whether there is moving water inrush in the image sequence, to record the specific number of water inrush in the background, and to count the number of moving water inrush parts at each moment;

[0061] A machine learning method based on a neural network of time series is used to predict the changes in the number of statistical water gushing parts. Training is performed based on the video surveillance data samples collected at the tunnel site. The corresponding threshold of the number of water gushing points in the video is set based on the training results. If the number of water gushing points increases at the next moment or at a certain moment after the current moment, that is, the increase in the number of water gushing parts in the video exceeds the threshold range, an early warning of the increase in the number of water gushing points will be issued.

[0062] The water inrush warning includes: using inter-frame difference to perform differential calculation on two adjacent frames or multiple adjacent frames in the image sequence, extracting the water inrush part in the tunnel monitoring video, and integrally calculating the contour and area of ​​the water inrush part; estimating the amount of water inrush based on the water inrush contour area calculated above, approximately equating the water inrush flow to a cylinder, and approximately obtaining the size of the water inrush volume; performing feature extraction, feature matching, geometric calculation and volume measurement calculation on the pre-processed video image, and the specific implementation is as follows:

[0063] Feature extraction is to select "edge" and "contour" as the features of the sudden water to be detected in the pre-processed video image, and use the Canny edge detection algorithm to extract the contour of the sudden water;

[0064] Feature matching is to extract the contour of the sudden water flow to be measured, match it with the calibration object of known size in the tunnel, establish the relationship between the sudden water flow to be measured and the known size, so as to realize the measurement of the sudden water flow contour;

[0065] The geometric calculation is to calculate the actual size of the sudden water to be measured by integration based on the proportional relationship between objects after feature matching. The pixel distance between feature points or feature contours can be converted into actual length or area size through geometric calculation. Finally, the actual area of ​​the sudden water is obtained according to the pixel distance between feature points and geometric calculation. The sudden water flow is approximately equivalent to a cylinder to approximately obtain the size of the sudden water volume.

[0066] The early warning mode method for water inrush volume is similar to the early warning mode for the increase in the number of water inrush points mentioned above. The water inrush volume is calculated and statistically analyzed based on time, and the water inrush volume is predicted by a machine learning method based on a time series neural network. Video surveillance data samples collected at the tunnel site are used for training, and corresponding thresholds are set based on the training results. If the water inrush volume increases at the next moment or at a certain moment after the current moment, that is, the increase in the water inrush part in the video exceeds the threshold range, an early warning of increased water inrush volume is issued.

[0067] Early warning of water quality changes in springs includes: water quality identification based on the different color depths of springs with different water qualities when reflected in the monitoring video. For the spring targets extracted by the frame difference method, the pixel difference detection algorithm in computer vision technology is used to check whether the color of some areas in the video has changed. If the difference exceeds the preset threshold, it is considered that the color has changed significantly, and the corresponding water quality has changed. The specific implementation is as follows:

[0068] The water-gushing area obtained by the frame difference method is extracted separately, and the pixel difference detection algorithm is implemented using the OpenCV library to extract the color information of the water-gushing area in each frame of the image; the color change metric is calculated, and the color information extracted in each frame is compared with the reference frame (clear water flow) to calculate the color difference metric, such as the number of different color pixels, color histogram difference, etc.

[0069] The machine learning algorithm is trained through a large number of surveillance videos collected on-site and indoor simulation experiments, and the color change threshold is set according to the machine training results. The color change of each frame of the video is measured and compared with the set threshold to determine whether the color change of the frame is obvious. If the difference exceeds the preset threshold, it is considered that the color has changed significantly, and the corresponding water quality has changed, and an early warning information on the change in water quality is issued.

[0070] In this implementation, preferably, the data fusion theory is used to fuse the above three types of indicators for early warning. According to the number of water gushing points, the size of water gushing volume and different level thresholds of sudden water quality, the hierarchical analysis method is used to determine the influence weights of the above three indicators on the water gushing level. The weights of each part are combined with expert evaluation to determine the water gushing state risk level score. The specific implementation is as follows:

[0071] The water inrush risk level score is as follows, where F is the threshold of three levels set for three types of indicators, and W is the influence weight of the above three types of indicators on the water inrush risk respectively established by the hierarchical analysis method. The sum of the two is the comprehensive evaluation index S of sudden water inrush risk:

[0072] F=[f 1 , f 2 , f 3 ] (3);

[0073] W=[ω 1 ,ω 2 ,ω 3 ] T (4);

[0074] S = F·ω (5);

[0075] According to the on-site video monitoring results or the test data of the sudden water burst video monitoring model, samples are constructed for training. The SVM support vector machine method in the machine learning algorithm is used for training to obtain the scores of the comprehensive evaluation index S under different degrees of water bursting conditions. The impact of the fusion of the three types of indicators on sudden water burst disasters is determined based on this, and the three thresholds are divided accordingly, namely low risk of water bursting, medium risk of water bursting, and high risk of water bursting.

[0076] This implementation method combines the above results to determine the warning indicators of different levels of sudden water disasters, issues sudden water disaster warnings in tunnels, and uses machine learning and deep learning methods to train based on existing videos to obtain accurate thresholds and realize the warning function; the sudden water disaster video warning specifically uses machine learning and deep learning methods, combined with convolutional neural networks, so that the algorithm is continuously trained and learned for video monitoring, thereby realizing more accurate automatic identification and warning functions for sudden water disasters; in this implementation method, GUI programming in java web is also used to design human-computer interaction programs, and a tunnel disaster monitoring and early warning platform is established. Both web version and client version are used to enable monitoring personnel to view monitoring videos in real time from computers or mobile phones, view relevant data of each monitoring equipment in real time, and receive alarm information.

[0077] like Figure 2 , Figure 3 and Figure 4 As shown, in this implementation, the monitoring equipment is installed in the tunnel monitoring section, and the high-definition camera is fixed on the tunnel secondary lining arch frame trolley by using expansion screws. Looking straight ahead can ensure that the heading face, side walls and arch top in front of the excavation position can be clearly seen, so as to achieve a better monitoring effect on the occurrence of sudden water disasters.

[0078] Preferably, in view of the dark environment and the high dust in the tunnel, the high-definition camera has the function of accurate recognition in this environment. Specifically, the high-definition camera has the ability to work normally in a dark environment and take clear pictures. The chip of the high-definition camera is that its light board should have an array infrared light board, which is characterized in that it provides auxiliary light source for the lens in the dark environment of the tunnel, so that the lens can sense the infrared light and turn it into an image, and can shoot accurate videos of sudden water movement, so that the monitoring video can be analyzed in real time with algorithms later, and the automatic recognition and early warning functions of sudden water disasters can be realized.

[0079] The HD camera of this implementation should be able to work normally and shoot clearly in an environment with a lot of floating dust. Specifically, in the design of the HD camera, the protective glass of the lens is designed to be inclined at an angle of 10° to 30° with the camera. Through mechanical analysis, this design can effectively prevent the adsorption and accumulation of dust on the camera glass, so that even if the camera is in a high dust environment for a long time, the camera will not be hindered by the accumulation of dust. At the same time, through optical analysis, it can be found that the reflection of light under the inclination of the glass within this range has little effect on the clarity of the video shot by the camera, and the clarity of the shooting can also be guaranteed, and the reflection or unclear shooting will not be caused by the reflection of light. The front glass lens of the HD camera of this implementation is made of heat-resistant glass, with a surface waterproof coating on the outside, and a circle of heating wire is set on the side to ensure that the fog attached to the lens can be quickly removed in a tunnel with heavy water vapor, ensuring the shooting accuracy.

[0080] The high-definition camera described in this implementation is fixed on a lens mount, and the lens mount is firmly fixed to the tunnel secondary lining trolley or the tunnel side wall through expansion screws, so as to monitor for a long time without affecting other work in the tunnel.

[0081] like Figure 3 As shown, the function of the high-definition camera top cover 1 is to prevent dust in the tunnel from affecting the lens; the lens protection lens 2 is tilted to effectively prevent dust in the tunnel air from accumulating on the lens and affecting the shooting effect; the lens protection lens 2 is arranged on the outside of the high-performance camera lens 3; the array-type infrared fill light board 4 is used to provide supplementary light source in the dark environment of the tunnel; the high-definition camera body 8 is fixed on the camera mounting base plate 6 through the camera angle adjustment device 5, and the camera mounting base plate 6 is used to fix the high-definition camera body 8 on the trolley; the heating wire 7 prevents the lens from easily generating water mist in the water-gushing tunnel, thereby ensuring the clarity of the shooting.

[0082] In this implementation method, the video data collected by the camera is processed by an algorithm and uploaded (using 4G or 5G network or local area network) through a data acquisition box (with a built-in control terminal), and uploaded to the corresponding tunnel disaster monitoring and early warning platform or client, where it is identified accordingly. If a water inrush situation meets the requirements of real-time monitoring and early warning of emergencies, it can be viewed in real time by the monitoring personnel.

[0083] The present invention studies the occurrence and positioning of sudden water inrush disasters in tunnels through the method of video monitoring and early warning, proposes corresponding equipment to collect information such as water inrush points, water inrush volume, and water quality in the tunnel, and analyzes and processes it through machine vision methods to determine whether a sudden water inrush disaster occurs in the tunnel, as well as information such as the size of the disaster, and comprehensively determines different degrees of early warning, which greatly reduces construction risks, provides a strong basis and guarantee for safe construction, and effectively avoids sudden water inrush disasters during tunnel construction.

[0084] The present invention can realize automatic identification of sudden water disasters, solve the problem that human participation is required in traditional sudden water monitoring, and overcome the extremely dangerous nature of the sudden occurrence of disasters for monitoring personnel compared with traditional manual monitoring. Compared with conventional manual video monitoring, the invention adds a machine vision algorithm, which solves the problem that people need to keep staring at the video during conventional video monitoring, realizes real-time monitoring and early warning of sudden water disasters, realizes high-definition unmanned monitoring and early warning in relatively poor environments such as dim dust in tunnels, combines machine vision and deep learning neural networks, and more accurately realizes accurate identification and early warning of sudden water disasters, and the camera of the present invention can still have the function of high-definition video recording in dim dusty and humid environments.

[0085] Embodiment 2:

[0086] like Figure 5 As shown, this implementation provides a tunnel water inrush disaster automatic identification and early warning system using a frame difference method, including:

[0087] The image processing unit is configured to: obtain images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binarized image, perform connectivity processing on the binarized image to obtain a target image of a moving target after excluding intrusion by a worker or a mobile device, and thereby obtain a target image sequence in the video;

[0088] The water gushing point quantity warning unit is configured to: obtain the number of water gushing points at the current moment and multiple moments before the current moment of the target image sequence, use a neural network model based on time series to predict the change of the number of water gushing points, and when the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, issue a warning that the number of water gushing points increases;

[0089] The water gushing volume warning unit is configured to: integrally calculate the contour and area of ​​the sudden water gushing according to the target image, estimate the amount of water gushing based on the calculated contour and area of ​​the sudden water gushing, equate the gushing water flow to a cylinder to obtain the sudden water gushing volume, obtain the sudden water gushing volume at the current moment and multiple moments before the current moment of the target image sequence, use a neural network model based on time series to predict the sudden water gushing volume change, and when the sudden water gushing volume at a certain moment after the current moment is greater than a second set threshold, issue a warning of the increase of the sudden water gushing volume;

[0090] The water quality early warning unit for gushing water is configured to: extract the image of the gushing water area according to the target image, extract the color information of the gushing water area in each video frame, compare the color information extracted in each frame with the color information of the reference frame, calculate the color difference metric, obtain the color difference metric of the current moment and multiple moments before the current moment of each video frame, use a neural network model based on time series to predict the change of the color difference metric, and when the color difference metric at a certain moment after the current moment is greater than a third set threshold, issue an early warning of water quality change.

[0091] The comprehensive warning unit is configured to: conduct comprehensive warning based on the number of water inrush points, sudden water volume and water quality changes, and calculate the comprehensive warning score S, including: S = f 1 *ω 1 +f 2 *ω 2 +f 3 *ω 3 , where ω 1 ,ω 2 and ω 3 are weight parameters, f 1 is the first set threshold, f 2 is the second set threshold, f 3 A threshold is set for the third time, and a comprehensive warning index S of the current moment and multiple moments before the current moment of the target image sequence is obtained. The comprehensive warning score S is predicted using a neural network model based on a time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing, and high risk of water gushing.

[0092] It is understandable that the above-mentioned units can be separately or completely combined into one or several other units to constitute, or one (some) of the units can be further divided into multiple functionally smaller units to constitute, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present application. The above-mentioned units are divided based on logical functions. In practical applications, the functions of one unit can also be implemented by multiple units, or the functions of multiple units can be implemented by one unit. In other embodiments of the present application, the tunnel flood disaster frame difference method automatic identification system can also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented by the collaboration of multiple units.

[0093] According to another embodiment of the present application, the system described in this embodiment can be constructed, and the automatic identification and early warning method of tunnel water inrush disasters using the frame difference method of the embodiment of the present application can be implemented by running a computer program (including program code) capable of executing the steps involved in the corresponding method described in Example 1 on a general computing device such as a computer including processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, and loaded into the above-mentioned computing device through the computer-readable recording medium, and run therein.

[0094] Embodiment 3:

[0095] like Figure 6 As shown, this implementation provides an electronic device, which includes a processor 1001, a communication interface 1002, and a computer-readable storage medium 1003. The processor 1001, the communication interface 1002, and the computer-readable storage medium 1003 may be connected via a bus or other means.

[0096] Among them, the communication interface 1002 is used to receive and send data, the computer-readable storage medium 1003 can be stored in the memory of the electronic device, the computer-readable storage medium 1003 is used to store a computer program, the computer program includes program instructions, and the processor 1001 is used to execute the program instructions stored in the computer-readable storage medium 1003.

[0097] Processor 1001 (or CPU (Central Processing Unit)) is the computing core and control core of the electronic device, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to implement corresponding method processes or corresponding functions.

[0098] The processor 1001 is configured to execute the following process:

[0099] Acquire images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binary image, perform connectivity processing on the binary image to obtain a target image of a moving target after excluding intrusion by workers or mobile equipment, and thereby obtain a target image sequence in the video;

[0100] The number of water gushing points at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change in the number of water gushing points is predicted using a neural network model based on time series. When the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, an early warning of an increase in the number of water gushing points is issued.

[0101] According to the target image, integrally calculating the contour and area of ​​the sudden water surge, estimating the amount of the sudden water surge based on the calculated contour and area of ​​the sudden water surge, and converting the sudden water surge flow into a cylinder to obtain the sudden water surge volume;

[0102] The volume of sudden water surge at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change of the sudden water surge volume is predicted by using a neural network model based on time series. When the sudden water surge volume at a certain moment after the current moment is greater than a second set threshold, an early warning of the increase in the sudden water surge volume is issued.

[0103] Extracting a water gushing area image according to the target image, extracting color information of the water gushing area in each video frame, comparing the color information extracted in each frame with the color information of the reference frame, and calculating a color difference metric;

[0104] The color difference measurement of the current moment and multiple moments before the current moment of each video frame is obtained, and the change of the color difference measurement is predicted using a neural network model based on time series. When the color difference measurement at a certain moment after the current moment is greater than a third set threshold, an early warning of water quality change is issued.

[0105] The number of water inrush points, sudden water volume and water quality changes are used for comprehensive early warning, and the comprehensive early warning score S is calculated, including:

[0106] S=f 1 *ω 1 +f 2 *ω 2 +f 3 *ω 3 , where ω 1 ,ω 2 and ω 3 are weight parameters, f 1is the first set threshold, f 2 is the second set threshold, f 3 Setting a threshold for the third;

[0107] The comprehensive warning index S of the target image sequence at the current moment and multiple moments before the current moment is obtained, and the comprehensive warning score S is predicted using a neural network model based on time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing and high risk of water gushing.

[0108] Embodiment 4:

[0109] This implementation provides a computer-readable storage medium (Memory), which is a memory device in an electronic device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the electronic device and, of course, extended storage media supported by the electronic device. The computer-readable storage medium provides a storage space that stores the processing system of the electronic device.

[0110] In addition, the storage space also stores one or more instructions suitable for being loaded and executed by the processor, and these instructions may be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here may be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as at least one disk storage; optionally, it may also be at least one computer-readable storage medium located away from the aforementioned processor.

[0111] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to implement the following process:

[0112] Acquire images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binary image, perform connectivity processing on the binary image to obtain a target image of a moving target after excluding intrusion by workers or mobile equipment, and thereby obtain a target image sequence in the video;

[0113] The number of water gushing points at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change in the number of water gushing points is predicted using a neural network model based on time series. When the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, an early warning of an increase in the number of water gushing points is issued.

[0114] According to the target image, integrally calculating the contour and area of ​​the sudden water surge, estimating the amount of the sudden water surge based on the calculated contour and area of ​​the sudden water surge, and converting the sudden water surge flow into a cylinder to obtain the sudden water surge volume;

[0115] The volume of sudden water surge at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change of the sudden water surge volume is predicted by using a neural network model based on time series. When the sudden water surge volume at a certain moment after the current moment is greater than a second set threshold, an early warning of the increase in the sudden water surge volume is issued.

[0116] Extracting a water gushing area image according to the target image, extracting color information of the water gushing area in each video frame, comparing the color information extracted in each frame with the color information of the reference frame, and calculating a color difference metric;

[0117] The color difference measurement of the current moment and multiple moments before the current moment of each video frame is obtained, and the change of the color difference measurement is predicted using a neural network model based on time series. When the color difference measurement at a certain moment after the current moment is greater than a third set threshold, an early warning of water quality change is issued.

[0118] The number of water inrush points, sudden water volume and water quality changes are used for comprehensive early warning, and the comprehensive early warning score S is calculated, including:

[0119] S=f 1 *ω 1 +f 2 *ω 2 +f 3 *ω 3 , where ω 1 ,ω 2 and ω 3 are weight parameters, f 1 is the first set threshold, f 2 is the second set threshold, f 3 Setting a threshold for the third;

[0120] The comprehensive warning index S of the target image sequence at the current moment and multiple moments before the current moment is obtained, and the comprehensive warning score S is predicted using a neural network model based on time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing and high risk of water gushing.

[0121] Embodiment 5:

[0122] The present implementation provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the electronic device performs the following process:

[0123] Acquire images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binary image, perform connectivity processing on the binary image to obtain a target image of a moving target after excluding intrusion by workers or mobile equipment, and thereby obtain a target image sequence in the video;

[0124] The number of water gushing points at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change in the number of water gushing points is predicted using a neural network model based on time series. When the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, an early warning of an increase in the number of water gushing points is issued.

[0125] According to the target image, integrally calculating the contour and area of ​​the sudden water surge, estimating the amount of the sudden water surge based on the calculated contour and area of ​​the sudden water surge, and converting the sudden water surge flow into a cylinder to obtain the sudden water surge volume;

[0126] The volume of sudden water surge at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change of the sudden water surge volume is predicted by using a neural network model based on time series. When the sudden water surge volume at a certain moment after the current moment is greater than a second set threshold, an early warning of the increase in the sudden water surge volume is issued.

[0127] Extracting a water gushing area image according to the target image, extracting color information of the water gushing area in each video frame, comparing the color information extracted in each frame with the color information of the reference frame, and calculating a color difference metric;

[0128] The color difference measurement of the current moment and multiple moments before the current moment of each video frame is obtained, and the change of the color difference measurement is predicted using a neural network model based on time series. When the color difference measurement at a certain moment after the current moment is greater than a third set threshold, an early warning of water quality change is issued.

[0129] The number of water inrush points, sudden water volume and water quality changes are used for comprehensive early warning, and the comprehensive early warning score S is calculated, including:

[0130] S=f 1 *ω 1 +f 2 *ω 2 +f 3*ω 3 , where ω 1 ,ω 2 and ω 3 are weight parameters, f 1 is the first set threshold, f 2 is the second set threshold, f 3 Setting a threshold for the third;

[0131] The comprehensive warning index S of the target image sequence at the current moment and multiple moments before the current moment is obtained, and the comprehensive warning score S is predicted using a neural network model based on time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing and high risk of water gushing.

[0132] A person skilled in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0133] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loading and executing computer program instructions on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instruction can be stored in a computer-readable storage medium or transmitted by a computer-readable storage medium. The computer instruction can be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (for example, coaxial cable, optical fiber, digital line (DSL)) or wireless (for example, infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server, a data center, etc. that contains one or more available media integration. Available media can be magnetic media (for example, floppy disk, hard disk, tape), optical media (for example, DVD), or semiconductor media (for example, solid-state drive (Solid State Disk, SSD)) and the like.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A frame difference method for automatic identification and early warning of tunnel water inrush disasters, characterized in that: The process includes: Acquire images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binary image, perform connectivity processing on the binary image to obtain a target image of a moving target after excluding intrusion by workers or mobile equipment, and thereby obtain a target image sequence in the video; The number of water gushing points at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change in the number of water gushing points is predicted using a neural network model based on time series. When the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, an early warning of an increase in the number of water gushing points is issued.

2. The automatic identification and early warning method for tunnel water inrush disasters using frame difference method according to claim 1, characterized in that: According to the target image, integrally calculating the contour and area of ​​the sudden water surge, estimating the amount of the sudden water surge based on the calculated contour and area of ​​the sudden water surge, and converting the sudden water surge flow into a cylinder to obtain the sudden water surge volume; The volume of sudden water surge at the current moment and multiple moments before the current moment of the target image sequence is obtained, and the change of the sudden water surge volume is predicted by using a neural network model based on time series. When the sudden water surge volume at a certain moment after the current moment is greater than a second set threshold, an early warning of the increase in the sudden water surge volume is issued.

3. The automatic identification and early warning method for tunnel water inrush disasters using frame difference method according to claim 2, characterized in that: Extracting a water gushing area image according to the target image, extracting color information of the water gushing area in each video frame, comparing the color information extracted in each frame with the color information of the reference frame, and calculating a color difference metric; The color difference measurement of the current moment and multiple moments before the current moment of each video frame is obtained, and the change of the color difference measurement is predicted using a neural network model based on time series. When the color difference measurement at a certain moment after the current moment is greater than a third set threshold, an early warning of water quality change is issued.

4. The automatic identification and early warning method for tunnel water inrush disasters using frame difference method according to claim 3 is characterized in that: The number of water inrush points, sudden water volume and water quality changes are used for comprehensive early warning, and the comprehensive early warning score S is calculated, including: S=f1*ω1+f2*ω2+f3*ω3, where ω1, ω2 and ω3 are all weight parameters, f1 is the first set threshold, f2 is the second set threshold, and f3 is the third set threshold; The comprehensive warning index S of the target image sequence at the current moment and multiple moments before the current moment is obtained, and the comprehensive warning score S is predicted using a neural network model based on time series. According to the comprehensive warning score S, the warning level is divided into low risk of water gushing, medium risk of water gushing and high risk of water gushing.

5. A tunnel flood disaster automatic identification and early warning system using frame difference method, characterized in that: include: The image processing unit is configured to: obtain images of adjacent video frames, subtract grayscale values ​​of the adjacent video frames pixel by pixel to obtain a differential image, perform binarization processing on the pixels in the differential image to obtain a binarized image, perform connectivity processing on the binarized image to obtain a target image of a moving target after excluding intrusion by a worker or a mobile device, and thereby obtain a target image sequence in the video; The water gushing point quantity warning unit is configured to: obtain the number of water gushing points at the current moment of the target image sequence and multiple moments before the current moment, use a neural network model based on time series to predict the change in the number of water gushing points, and when the number of water gushing points at a certain moment after the current moment is greater than a first set threshold, issue a warning of an increase in the number of water gushing points.

6. The automatic identification and early warning system for tunnel water inrush disasters using frame difference method as claimed in claim 5, characterized in that: The water gushing volume warning unit is configured to: integrally calculate the contour and area of ​​the sudden water gushing according to the target image, estimate the amount of water gushing based on the calculated contour and area of ​​the sudden water gushing, equate the gushing water flow to a cylinder to obtain the sudden water gushing volume, obtain the sudden water gushing volume at the current moment and multiple moments before the current moment of the target image sequence, use a neural network model based on time series to predict the sudden water gushing volume change, and when the sudden water gushing volume at a certain moment after the current moment is greater than a second set threshold, issue a warning of the increase of the sudden water gushing volume; The gushing water quality early warning unit is configured to: extract a gushing water area image according to the target image, extract color information of the gushing water area in each video frame, compare the color information extracted in each frame with the color information of the reference frame, calculate the color difference metric, obtain the color difference metric of the current moment and multiple moments before the current moment of each video frame, use a neural network model based on time series to predict the change of the color difference metric, and issue an early warning of water quality change when the color difference metric at a certain moment after the current moment is greater than a third set threshold; The comprehensive warning unit is configured to: make a comprehensive warning based on the number of water gushing points, the volume of sudden water gushing, and the change in water quality, calculate the comprehensive warning score S, including: S=f1*ω1+f2*ω2+f3*ω3, wherein ω1, ω2, and ω3 are all weight parameters, f1 is the first set threshold, f2 is the second set threshold, and f3 is the third set threshold, obtain the comprehensive warning index S of the current moment of the target image sequence and multiple moments before the current moment, use a neural network model based on time series to predict the comprehensive warning score S, and divide the warning level into low risk of water gushing, medium risk of water gushing, and high risk of water gushing according to the comprehensive warning score S.

7. A tunnel flood disaster automatic identification and early warning system using frame difference method, characterized in that: include: A high-definition camera and a control terminal communicating with the high-definition camera, wherein the control terminal is configured to execute the automatic identification and early warning method for tunnel water inrush disasters using a frame difference method as described in any one of claims 1 to 4; The high-definition camera comprises: a high-definition camera body, a high-definition camera top cover, an inclined lens protection lens, a camera lens, an array-type infrared fill light board, a camera angle adjustment mechanism, a camera mounting bottom plate and a heating wire; The top cover of the HD camera is fixed on the top of the HD camera body, the inclined lens protection glass is arranged on the outside of the camera lens, the lens protection glass and the camera lens are arranged at the lower part of the top cover of the HD camera, an array-type infrared fill light board is arranged next to the camera lens, the HD camera body is connected to the camera mounting base plate through a camera angle adjustment mechanism, and heating wires are arranged around the lens protection glass.

8. A computer device, characterized in that: include: A processor and a computer readable storage medium; a processor adapted to execute a computer program; A computer-readable storage medium having a computer program stored therein, wherein when the computer program is executed by the processor, the method for automatically identifying and warning of tunnel water inrush disasters using the frame difference method as described in any one of claims 1 to 4 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor and executing the automatic identification and early warning method for tunnel water inrush disasters using the frame difference method as described in any one of claims 1 to 4.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the automatic identification and early warning method for tunnel water inrush disasters using the frame difference method as described in any one of claims 1 to 4 is implemented.

Images